1 Introduction

Hans Broekhuis* and Ralf Vogel**

ABSTRACT This chapter will motivate why it is useful to consider the topic of deri-vations and fi ltering in more detail. We will argue against the popular belief that the minimalist program and optimality theory are incompatible theories in that the former places the explanatory burden on the generative device (the computational system CHL) whereas the latter places it on the fi ltering device (the OT evaluator). Although this belief may be correct in as far as it describes existing tendencies, we will argue that minimalist and optimality theoretic approaches normally adopt more or less the same global architecture of grammar: both assume that a generator defi nes a set S of potentially well-formed expressions that can be generated on the basis of a given input and that there is an evaluator that selects the expressions from S that are actually grammatical in a given language L. For this reason, we believe that it has a high priority to investigate the role of the two components in more detail in the hope that this will provide a better understanding of the differences and simi-larities between the two approaches. We will conclude this introduction with a brief review of the studies collected in this book.

1. The architecture of grammar

The studies collected in this book all discuss the relation between the generative and the fi lter component of the grammar. The focus will be on syntax although the collec-tion also contains a contribution by John J. McCarthy and Kathryn Pruitt, which discusses the issue for phonology. The starting point of this book is the popular view that current generative theories differ considerably in where they place the burden of explanation: whereas minimalist approaches generally assume that this is the generative component (the computational system CHL), optimality-theoretic approaches generally focus on the fi lter component (the OT-evaluator). This differ-ence between the minimalist program (MP) and optimality theory (OT) is also refl ected in the claims that are normally made about the output of the generator;

* Hans Broekhuis, Meertens Institute, P.O.-box 94264, 1090 GG Amsterdam, The Netherlands. E-mail: hans.broekhuis@meertens.knaw.nl

** Ralf Vogel, Universität Bielefeld, Fakultät für Linguistik und Literaturwissenschaft, P.O.-Box 10 01 31, 33501 Bielefeld, Germany. E-mail: rvogel@uni-bielefeld.de)

2 Hans Broekhuis and Ralf Vogel

minimalist approaches normally presuppose that the output of CHL is small and may in fact be restricted to a single representation in many cases; optimality-theoretic approaches, on the other hand, normally maintain that the generator creates a candi-date set that is very large or even infi nite. It is important to note, however, that propo-nents of MP normally accept the idea that the generator may overgenerate and that we must therefore assume additional means to fi lter out the unwanted structures from the reference set. This means that many proponents of MP and OT do agree that the global architecture of grammar has the form in Figure 1.1, where the Generator and the Evaluator can be held responsible for respectively the universal and language-specifi c properties of languages. The essential property of this model is that the generator defi nes a set S of potentially well-formed expressions that can be gener-ated on the basis of a given input, and that the evaluator selects those expressions from S that are actually grammatical in a given language L.

This general id ea is, of course, not new and has already been formulated by Chomsky and Lasnik in ‘Filters and control’ (1977), where it is argued that ‘to attain explanatory adequacy it is in general necessary to restrict the class of possible gram-mars, whereas the pursuit of descriptive adequacy often seems to require elaborating the mechanisms available and thus extending the class of possible grammars’. In order to solve this tension they propose that ‘there is a theory of core grammar with highly restricted options, limited expressive power, and a few parameters’ next to a more peripheral system of ‘added properties of grammar’, which ‘we may think of as the syntactic analogue of irregular verbs’. Chomsky and Lasnik assume that core grammar consists of the phrase structure and transformational rules (the generator in Figure 1.1), whereas the more peripheral system consists of language-specifi c surface fi lters (the evaluator), and claim that the introduction of these fi lters contrib-utes to the simplifi cation of the transformational rules by bearing ‘the burden of accounting for constraints which, in the earlier and far richer theory, were expressed in statements of ordering and obligatoriness, as well as all contextual dependencies that cannot be formulated in the narrower framework of core grammar’.

The ideas about which aspects of grammar should be considered part of core grammar or part of the periphery have, of course, considerably changed over the years; the that-trace fi lter, for example, was originally proposed as a language-specifi c fi lter for English, but the Empty Category Principle, which ultimately grew out of it, was assumed to be part of core grammar. Nevertheless, the gist of the proposal has survived in the more recent minimalist incarnations of the theory, where core syntax can be more or less equated with CHL, and the periphery with the inter-face conditions. The task of reducing core grammar as much as possible has been very successful: the reduction of CHL to its absolute minimum (internal and external merge) much contributes to the explanatory adequacy of the theory in the technical

Input EvaluatorOptimaloutput

OutputrepresentationsGenerator

Figure 1.1 The architecture of grammar.

Introduction 3

sense that it provides a principled basis, independent of any particular language, for the selection of the descriptively adequate grammar of each language (although it still remains to be shown that the minimalist conception of UG is more successful in core explanatory tasks like modeling language acquisition, explaining the origins of language and so on, than its predecessors). But, as expected, the contribution of core grammar to the descriptive adequacy of the theory has diminished accordingly, so that in this respect we have to rely more and more on the interface conditions.

That the global architecture of grammar has been like that indicated in Figure 1.1 for over three decades now may have been obscured by the fact that earlier phases of the theory assumed a so-called T- or inverse Y-model, according to which the derivation of LF- and PF-representations diverge after a certain point (s-struc-ture or Spell-Out) in order to account for certain mismatches between linear order and semantic interpretation by deriving different input structures for the LF- and PF-component of the grammar. We will see in Section 2.1 that this property of the early principles-and parameters (P&P) models has disappeared in the later versions of MP and that, as a result, these later versions fully accord with the linear model in Figure 1.1. Now we have established this, we will briefl y discuss the form and the role of the fi lter component in various stages of the P&P framework, as well as in OT.

2. The fi lter component

Although Chomky and Lasnik originall y assumed that the fi lter component was not part of core grammar but of a language-specifi c periphery, it soon became clear that the fi lter component also had certain universal properties. For this reason, we will replace the notion of core grammar for the derivational component by the notions generator, computational system and narrow syntax. In order to avoid unwanted connotation we will likewise avoid the notion of periphery and use instead the notion of evaluator or fi lter component. The generator and the evaluator should both be considered part of core grammar.

2.1 The principles-and parameters approach

The introduction of a fi lter component in ‘Filters and control’ was motivated by the fact that this made a more restrictive formulation of narrow syntax possible by eliminating ordering statements and language-specifi c properties from the transfor-mational component of the grammar. By way of demonstration we will consider the derivation of the relative clauses in (1).

(1) (a) the man who I know (b) the man that I know (c) the man I know (d) *the man who that I know

4 Hans Broekhuis and Ralf Vogel

The relative pronoun who is base-generated in the regular object position, so that the d-structure of the examples in (1) is as given in (2a). Chomsky and Lasnik further proposed that universal grammar (UG) contains a universal principle ‘Move wh-phrase’ that requires that relative pronouns (and other wh-phrases) be placed to the left of the complementizer, as in the s-structure representation in (2b).

(2) (a) the man [that I know who] (d-structure) (b) the man [[CO MP who that] I know twho] (s-structure)

The examples in (1) can now be derived by assuming that UG contains a PF-rule Deletion that precedes the fi lters and freely deletes the relative pronoun who or the complementizer that; cf. Chomsky and Lasnik (1977: ex. (6)). The resulting PF-representations are given in (3). The desired grammaticality pattern is derived by postulation of the language-specifi c Doubly Filled COMP Filter, which prohibits the simultaneous realization of the relative pronoun and the complementizer in English. This excludes representation (3d).

(3) (a) the man [[COMP who that] I know twho] (b) the man [[COMP who that] I know twho] (c) the man [[COMP who that] I know twho] (d) *the man [[COMP who that] I know twho]

Although the deletion rule is freely applicable, the resulting representation is subject to a recoverability principle, which requires that deleted elements be locally recoverable. This is needed to block deletion of the wh-phrase in representations like (4): the recoverability principle in tandem with the Doubly Filled COMP Filter ensures that the representations in (4b–d) are excluded.

(4) (a) I wonder [who that you met twho] (b) *I wonder [who that you met twho] (c) *I wonder [who that you met twho] (d) *I wonder [who that you met twho]

By the same means, deletion of a preposed PP in relative clauses like (5) is blocked. Deletion of about which would violate the recoverability principle because the prep-osition about cannot be recovered locally.

(5) (a) the book [about which that he spoke tabout which] (b) *th e book [about which that he spoke tabout which] (c) *the book [about which that he spoke tabout which] (d) *the book [about which that he spoke tabout which]

The virtue of Chomsky and Lasnik’s proposal is that by accounting for the language-particular properties of English relative constructions by means of the Doubly Filled COMP Filter, we can keep the transformational rule that derives s-structure (2b) maximally simple (Move wh-phrase), which, in turn, makes it possible to attribute this rule to UG.

Introduction 5

In the Government and Binding (Chomsky 1981) and Barriers (Chomsky 1986) period, the attempts to further reduce the transformational component of narrow syntax led to the formulation of the general rule Move α. As far as the fi lter compo-nent was concerned, it turned out that some of the language-specifi c fi lters proposed in Chomsky and Lasnik (1977) had a wider application and could be reformulated as more general principles. For example, the so-called that-trace fi lter, which prohibits a trace immediately to the right of the complementizer that, was reformulated as/reduced to the Empty Category Principle (ECP), which requires that a trace be prop-erly governed. Although the ECP was claimed to be universal, that is, to be part of UG, its function is more or less the same as that of the that-trace fi lter: it excludes structures that have been created by narrow syntax. Therefore the formulation of the ECP is not a reason to frown with a skeptical eye on the notion of fi lter; it rather opened the prospect of obtaining a certain degree of explanatory adequacy in the domain of fi lters, so that the fi lter component could also enter the domain of core grammar.

In the Minimalist Program, as developed by Chomsky since the mid-1980s, the generator seems to have been reduced to its absolute minimum. The computa-tional system of human language CHL, as it is now called, consists essentially of one merge operation in two guises. External merge combines two independent syntactic objects into a larger syntactic unit, whereas internal merge takes some element from an existing syntactic object, and merges it to the root of this object, thus deriving the effect of movement. Merge is subject to a number of general conditions. For example, it never involves more than two objects at the same time, which results in binary branching phrase structures. Internal Merge obeys certain locality restrictions and is further subject to the Last Resort Condition, which requires that movement be triggered by some unvalued formal feature. As in Chomsky and Lasnik (1977), descriptive adequacy lies mainly outside the computational system: Chomsky (1995: §4.7.3), for example, suggests (rightly or wrongly) that ‘rearrangement’ phenomena like extraposition, right-node raising, VP-adjunction and scrambling are essentially the result of stylistic rules of the phonological component.

Many of the fi lters as discussed in Chomsky and Lasnik (1977) have not found an alternative account in MP, but the fact that they are not discussed is, of course, no guarantee that they are not needed. In this connection it is important to note that Chomsky (1995) explicitly claims that CHL generates a set of converging (= poten-tially well-formed) derivations satisfying Full Interpretation, the so-called refer-ence set, from which the admissible structures are selected by a number of global economy conditions: derivations with a smaller number of derivational steps are preferred (fewest steps), as are derivations with shorter movement chains (shortest steps).

The language L thus generates three relevant sets of derivations: the set D of derivations, a subset DC of convergent derivations of D, and a subset DA of admissible derivations of D. FI determines DC, and the economy conditions select DA. … DA is a subset of DC. (Chomsky 1995: 220)

6 Hans Broekhuis and Ralf Vogel

It is not so clear whether global economy conditions still play a role in the current versions of MP. It seems that very soon they lost independent status by being incor-porated into the defi nition of the movement operation: fewest steps was replaced by Last Resort (Chomsky 1995: 280) and shortest steps by the Phase Impenetrability Condition proposed in Chomsky (2001). As a result, DC and DA can be considered identical and we are left with only two sets of derivations: the set of derivations D and the set of converging derivations DC. Proponents of the so-called crash-proof syntax framework claim that nothing more is needed; more specifi cally they claim that ‘no fi lters are imposed on the end products of derivations, and no global fi lters (e.g. comparison of derivations) assign status to derivations as a whole’; see Frampton and Gutmann (2002: 90) and the contributions in Putnam (2010) for exten-sive discussions of the viability of this claim. Chomsky (1995: ch. 4, 221), however, maintained the more traditional line of thinking by introducing bare output condi-tions, which are later referred to as interface conditions, which are ‘imposed from the outside’ by the performance systems that make use of the representations created by CHL: the articulatory-perceptual and the conceptual-intentional system. Chomsky further claims that the interface conditions are involved in the displacement property of language and we will see below that he formulates these conditions in later work in the format of a fi lter on the output of CHL; cf. Chomsky (2001) and the discussion of (10/20) below.

We already noted that the early P&P models diverge from the linear model in Figure 1.1 in that the derivation of the PF- and LF-representations split at a certain point in the derivation in order to account by means of covert movement for the fact that there can be certain mismatches between linear order and semantic interpreta-tion. Very early in the development of MP, proposals have been put forth to eliminate this property from the grammar. Groat and O’Neil (1996), for example, show that the copy theory of movement makes it possible to account for th e discrepancies in PF and LF-representations by assuming that phonology can spell out either the lower or the higher copy in a movement chain; see also Bobaljik (2002). Chomsky (1995: ch. 4) argues that economy considerations can also account for these mismatches when we assume that it is more economical to move a syntactic category without its phonological features; pied piping of the phonological features is possible only when there are independent reasons to do so. The most recent development is the introduction of Agree (feature valuing at a distance) in the Minimalist Inquiry frame-work, which has made movement totally superfl uous from a computational point of view. These proposals have in common that they make it possible to assume that the derivation of the LF- and PF-representations proceed in fully parallel fashion. The model of the Minimalist Inquiry framework, for example, is therefore as indi-cated in Figure 1.2.

Since Agree makes movement superfl uous in the sense that it is no longer needed for feature checking, movement must be forced by other factors. More specifi cally, although movement must still be formally licensed by unvalued formal features, the question whether it actually applies depends on the interface conditions imposed by the conceptual-intentional (LF) or the articulatory-perceptual (PF) component on the output representations of CHL. The intuition underlying this proposal is actu-

Introduction 7

ally much older than the Minimalist Inquiry framework. For example, it has been argued that the motivation for wh-movement is that a wh-phrase can only be inter-preted if it heads an operator-variable chain; see for example Chomsky (1991: 440) and Rizzi (1996). What is new is that Chomsky (2001) claims that certain types of A-movement are also externally motivated. We will look at this in some detail in the remainder of this subsection.

According to MP, movement of a syntactic object S is subject to last resort: it must be triggered by some unvalued formal feature of a higher functional head H that can be checked or valued by a corresponding feature of S. In the earliest proposal it was assumed that these features of H come in two forms: weak and strong features. A strong feature on H must be checked before the projection of H is merged with some higher head; if checking does not take place, the derivation is canceled. A weak feature on H, on the other hand, cannot be checked before Spell-Out as a result of the economy condition Procrastinate. This proposal led to a very rigid system in which the question whether a certain movement does or does not apply is mechani-cally determined by the feature constellation of the functional head H. However, it is clear that movement may be sensitive to other factors as well. Consider the case of so-called object shift (OS) in the Icelandic examples in (6).

(6) (a) Jón ke ypti ekki bókina. bókina focus Jón bought not the book (b) Jón keypti bókinai ekki ti bókina presupposition

The examples in (6) demonstrate that it is possible in Icelandic to move the direct object to the left, across the negative adverb ekki. This movement is not obligatory, however, but depends on the information structure of the clause: OS applies only when the object is part of the presupposition (‘old’ information) of the clause; it is excluded when it is part of the focus (‘new’ information) of the clause.

Let us provisionally assume that OS is triggered by the case feature on the light verb v* (Vikner 1994; Chomsky 2001): if this case feature were strong, we would wrongly expect this movement to be obligatory; if it were weak, we would wrongly predict it to be impossible. In order to account for the apparent optionality of OS, we must therefore introduce additional means. One possibility would be to make the strength of the case feature sensitive to the information structure of the clause: only when the object is part of the presupposition of the clause does v* have a strong case feature. Apart from being ad hoc, this option is not descriptively adequate since OS is never possible in complex tense constructions like (7): OS is excluded irre-spective of the information structure of the clause, and (7a) is therefore ambiguous.

CHLInterface

ConditionsOptimaloutput

Output representations

Input

Figu re 1. 2 The Minimalist Inquiry model (Chomsky 2000 and later).

8 Hans Broekhuis and Ralf Vogel

(7) (a) Jón hefur ek ki keypt bókina. ambiguous Jón has not bought the book (b) *Jón hefur bókina ekki keypt tbokina —

Another possibility is to follow Holmberg (1999) in claiming that OS is actually not part of narrow syntax. He proposes that OS is a phonological (or, at least, post-spell out) operation that is driven by the interpretation of the object: in the termi-nology used above, OS is only possible if the object is part of the presupposition of the clause. This is stated in (8a), which paraphrases Chomsky’s (2001: (54a)) summary of Holmberg’s claim. Holmberg (1999: 22) accounts for the ungrammati-cality of (7b) by postulating the additional restriction on the application of OS in (8b): OS is blocked in (7b) because it would move the object across the main verb.

(8) (a) Object shift is a phonological movement that satisfi es condition (8b) and is driven by the semantic interpretation INT of the shifted object:

(i) INT: object is part of the presupposition of the clause. (ii) INT: object is part of the focus of the clause. (b) Object shift cannot apply across a phonologically visible category

asymmetrically c-commanding the object position except adjuncts.

Chomsky (2001: 32) argues that Holmberg’s proposal is problematic because ‘displacement rules interspersed in the phonological component should have little semantic effect’ (p. 15), and he therefore develops a proposal according to which OS takes place in narrow syntax. The relevant confi guration is given in (9), where Obj is the θ-position of the object, and XP is a specifi er position of v* created by OS (note that Chomsky assumes a multiple specifi er approach).

(9) … [αXP [Subject v* [ V … Obj ]]]

The representation in (9) is an intermediate stage in the derivation: at some later stage in the derivation the subject is moved into SpecTP and in simple tense constructions the v*+V complex is moved to T. Given this, Chomsky (2001: 61) tries to account for the properties of Icelandic OS in (8) by adopting the assumptions in (10), where INT and INT are again interpreted as in (8a).

(10) (a) v* is assigned an EPP-fea t ure only if that has an effect on outcome. (b) The EPP position of v* is assigned INT. (c) At the phonological border of v*P, XP is assigned INT.

The EPP-feature mentioned in (10a) has the same function as the strong features in the earlier proposals in the sense that it forces movement of some element into a specifi er position of the head that it is assigned to, but it is no longer considered an inherent lexical property of the lexical items. Instead, v* can in principle be freely assigned an EPP-feature and it is the function of the clause in (10a), which is claimed to be an invariant principle of grammar, to determine when this leads to an accept-able result; assignment of an EPP-feature to v* is only possible if the resulting move-

Introduction 9

ment has some effect on the output representation. According to Chomsky this is only the case when the movement affects the semantic/pragmatic interpretation of the clause, or when it makes A-movement possible (by placing the object at the phonological edge of the v*P-phase). We will see shortly that this leads to a less rigid system in the sense that movement can be made sensitive to factors other than the feature constellation of the attracting head.

Chomsky claims that (10b) is also an invariant principle: in the terminology employed earlier, this claim expresses that an object occupying the position XP in (9) in the output representation must be construed as being part of the presupposi-tion of the clause; see (8a). It is important to note that (10b) is only concerned with shifted objects, and leaves open the option that non-shifted objects are ambiguously interpreted as being part of either the focus or the presupposition of the clause. This is needed in order to allow the non-shifted objects in Icelandic examples like (7a) to be interpreted as part of the presupposition of the clause, and, of course, also correctly predicts that objects in languages like English, which do not have OS of the Icelandic sort, can be part of either the focus or the presupposition of the clause.

Given that (10b) does not restrict the interpretation of non-shifted objects, we need something in addition to account for the fact that OS is obligatory in examples like (6b). This is where (10c) comes in. Let us fi rst consider the notion of phonolog-ical border, which is defi ned as in (11); since Chomsky does not specify the notion of phonological material, we take it to refer to an abstract set of phonological features that will be spelled out in the PF-component.

(11) XP is at the phonological bord er of v*P, iff: (a) XP is a v*P-internal position, and; (b) XP is not c-commanded by v*P-internal phonological material.

The main difference between the examples in (6) and (7) is that in the former the main verb has moved out of v*P into T, whereas in (7) it has not and thus occupies a v*P-internal position. Example (7a) is therefore correctly predicted to be ambiguous: since the v*+V complex is v*P-internal and c-commands the object, clause (10c) does not apply and the object can be interpreted either as part of the focus of the clause (INT) or as part of the presupposition of the clause (INT). Example (7b) is consequently blocked by (10a) because OS has no effect on the outcome as the object can also be assigned the interpretation INT in its base position in (7a). Therefore, in constructions like (7), the EPP-feature can only be assigned to v* if it is needed to enable A-movement. In (6), on the other hand, there is no v*P-internal phonological material that c-commands the position Obj. Consequently, if the object occupies this position, (10c) states that it must be assigned INT. Movement of the object into the XP-position in (9) therefore has an effect on the outcome by licensing the interpreta-tion INT, and (10a) consequently allows assignment of an EPP-feature to v*.

It is important to note that statement (10c) clearly functions as a fi lter in the sense of Chomsky and Lasnik (1977). First, it is clear that it cannot be considered a condition on the derivation: when we would apply it to the intermediate stage in (9), the desired distinction between (6) and (7) could not yet be made locally, because the verb and the subject are moved out of the v*P only at a later stage in

10 Hans Broekhuis and Ralf Vogel

the derivation; Chomsky therefore assumes that it applies at the higher phase level (CP). Second, (10c) is a language-specifi c statement: Icelandic (and the continental Germanic languages) is subject to it, and therefore OS is forced in examples like (6b); the Romance languages, on the other hand, are not subject to it, so that (10a) blocks OS in comparable Romance examples. Thus, statement (10c) has the two characteristic properties of the PF-fi lters proposed in Chomsky and Lasnik (1977).

This subsection has shown that all grammars proposed during the P&P era have the global architecture of grammar indicated in Figure 1.1, although this was obscured in the early period by the assumption that derivations of the PF- and LF-representation diverge at some point in the derivation. It has been shown that by dropping this assumption Chomsky’s recent Minimalist Inquiry framework fully conforms to the architecture in Figure 1.1; the grammar consists of a generative component that creates representations that are subsequently evaluated by a fi lter component. The fi lters place both semantic and phonological constraints on the output of CHL, which refl ects the fact that the representation(s) that pass these fi lters are subsequently fed to the articulatory-perceptual and the conceptual-intentional system where they undergo further computation in order to receive, respectively, a phonetic and a semantic interpretation.

2.2 Optimality theory

Optimalit y theory fi ts nicely to the global architecture of grammar in Figure 1.1, which is clear from the fact that it can be found in virtually all introductory texts on OT. It therefore also fi ts in the generative tradition as described in Section 2.1, but crucially differs from the P&P framework in that the evaluator is not taken to consist of universal principles and language-specifi c fi lters. The guiding intuition is instead that such principles and fi lters can be more adequately expressed by means of the ranking of a set of more primitive violable constraints; see Figure 1.3. We refer the reader to Pesetsky (1997; 1998) and Dekkers (1999) for early demonstrations of this.

Furthermore, OT adopts a holistic conception of language in the sense that the grammaticality of an expression E for some language L cannot be established by inspecting E alone, but is determined by comparing it to other expressions produced by the generator. This normally seems to go far beyond what is discussed under the term transderivationality in early minimalism; a derivation that is blocked by an economy constraint yields an ungrammatical expression in minimalism, whereas a loser in one OT-competition may still be the winner of another competition. It must be noted, however, that the latter is also a property of the set of statements in (10), which shows that Chomsky’s most recent version of MP converges with this aspect of OT.

Figure 1.3 Optimality theo ry.

Input Rankedconstraints

Optimaloutput

Output representations

Generator

Introduction 11

Like the MP model in Figure 1.2, the OT model in Figure 1.3 entails two notions of well-formedness: one with respect to the generator and one with respect to the evaluator. The usual version of the generator in OT is, however, more liberal and unrestricted than CHL, and allows for a comparatively large candidate set. The OT-evaluator, of course, differs from the one proposed in MP in that it uses ranked constraints instead of interface condition, but they do resemble them in that they often incorporates aspects of the interpretative systems; cf. Vogel (2004; this volume).

An important dif ference between OT and MP is that the former can also be seen as a meta-theory or a methodological guideline; this is clear from the fact that it may be applied to a wide variety of empirical domains: it can be equally well applied to phonology as to, for example, syntax, and it is certainly conceivable that it can also be successfully applied outside the domain of linguistics. When we restrict ourselves to a certain empirical domain, it may be that the differences between the different OT-approaches are so small that it is actually justifi ed to speak of a more or less coherent theory. This might well be the case for OT-phonology, given that there seems to be considerable agreement among OT-phonologists on the nature of the input, the operations that can be performed by the generator, and the nature of the output. Furthermore, OT-phonologists do not only agree on the basic assump-tion that the evaluator consists of ranked violable constraints, but they also seem to share the belief that the postulated constraints are of just two types, the so-called faithfulness and markedness constraints. And, fi nally, there even seems be some consensus about the individual constraints that are needed. Of course, there are also hotly debated issues, such as the question of whether the constraints are part of an innate, universally available set CON, or whether they are acquired on the basis of the primary linguistic data.

The situation in OT-syntax is entirely different: we are clearly not dealing with a generally accepted theory. Although Figure 1.3 is very specifi c about the nature of the evaluator, which has the defi ning property of consisting of ranked violable constraints, the nature of the generator is left open entirely; the generator can take the form of virtually any imaginable generative device, and, as a result, the genera-tors of the current OT-approaches to syntax are based on different and often incom-patible linguistic theories. Some more or less random examples are given in (12).

(12) (a) Lexical-Functional Grammar : Bresnan (2000); Sells (2001) (b) Early Principles-and-Parameters Theory: Grimshaw (1997); Pesetsky (1998) (c) Minimalism: Dekkers (1999); Woolford (2007); Broekhuis (2008) (d) Others: Müller (2000/2001); Vogel (2006)

Since the generators postulated by the proposals in (12) differ considerably and the generated candidate sets will therefore be constituted by candidates with entirely different properties, the postulated constraints will be quite different as well. As a result, we are dealing with OT-approaches that are as different as (or perhaps even more different than) the theories on which the generator is modeled. We will illus-trate this below by comparing the OT-approaches proposed in Grimshaw (1997), Dekkers (1999) and Broekhuis (2008), which are all based on some version of the principles-and-parameters theory.

12 Hans Broekhuis and Ralf Vogel

Grimshaw’s (1997) proposal was originally written in the early 1990s and is based on the pre-minimalist principles-and-parameters framework. Among other things, this is clear from the fact that she tries to capture the directionality parameter, which was still generally assumed at that time, by means of two confl icting constraints HEAD LEFT and HEAD RIGHT (the head is leftmost/rightmost in its projection). In addition, she assumes the constraints SPECIFIER LEFT and SPECIFIER RIGHT (the specifi er is leftmost/rightmost in its projection). Given that Grimshaw also assumes that the structures created by the generator conform the general X-bar-schema, the linearization of these structures follows from the language-specifi c ranking of these four constraints. Broekhuis (2008), which is based on the minimalist machinery proposed in Chomsky (2000) and later work, need not make use of Grimshaw’s alignment constraints given that he adopts some version of Kayne’s (1994) Linear Correspondence Axiom, according to which linear order is derived from the hierarchical relation between the constituent in the output representation. In his approach, linear order therefore follows from the language-specifi c ranking of a set of so-called EPP-constraints, which favor movement of a goal into its probe’s minimal domain (in the sense of Chomsky 1995: ch. 3), and the economy constraint *MOVE, which disfavors movement. For example, the ‘strong’ ranking EPP(case) >> *MOVE requires movement of the probed noun phrase into the minimal domain of the unvalued case-features of v* or the infl ectional node I, whereas the ‘weak’ ranking *MOVE >> EPP(case) requires that the probe remain in its original position. Such EPP-constraints, which are used to express the same intuition as Chomsky’s Agree-based approach that Agree is normally suffi -cient for convergence, will fi nd no place in OT-approaches that follow Groat and O’Neil (1996) in assuming that feature checking invariably triggers movement and that the linear order depends on the question whether it is the tail or the head of the resulting chain that is spelled out; such approaches will replace the EPP-constraints, for example, by Dekker’s (1999) PARSE-F constraints, which favor pronunciation of moved constituents in the position of their formal features (the head of the chain), and reinterpret *MOVE as a constraint that favors pronunciation of moved elements in their base position (the tail of the chain).

The previous paragraph has shown that properties of the proposed generator are immediately refl ected in the nature of the postulated violable constraints of the OT-evaluator. The differences between the three OT-approaches discussed above are still relatively small due to the fact that the proposed generators all fi nd their origin in the Chomskyan generative tradition, but it will be clear that the differences between these OT-approaches and OT-approaches that are based on other (generative) tradi-tions may be much larger. For example, Broekhuis (2008) and Sells (2001) both develop an OT-analysis of Scandinavian object shift, but the two proposals differ at least as much as the minimalist and Lexical-Functional approaches that they are based on: whereas Broekhuis’ analysis is built on the restrictions on movement of the clausal constituents, Sells’ analysis is based on the restrictions on their phono-logical alignment.

Let us return to the guiding intuitions that connect all work in OT. The generator is an overgenerating system, which creates the candidate set from which the evalu-ator selects the optimal candidate(s) for a certain language L. The candidate set is

Introduction 13

generally assumed to be infi nite and to contain many candidates that will never surface because they are harmonically bound by some other candidate (where A is harmonically bound by B if A violates at least one constraint on top of the constraints violated by B). Furthermore, the focus of attention is on the evaluator, which consists of a set of constraints with the properties in (13a–c), which we will more extensively discuss below.

(13) The optimality theoretic e valuator contains constraints that: (a) are taken from a universal set of constraints CON; (b) are violable; and (c) have a language-specifi c ranking.

The constraints crucially differ from the language-specifi c fi lters assumed in the principles-and-parameters theories in that they are generally assumed to be universal, that is, to be part of a universal set of constraints CON. These constraints can express language-specifi c properties by virtue of their properties in (13b) and (13c): languages may differ in the ranking of the universal constraints, and thereby select different candidates as optimal as a result of the fact that violation of a lower ranked constraint is tolerated in order to satisfy a higher ranked constraint. The way the OT-evaluator works can readily be demonstrated by means of Pesetsky’s (1997; 1998) analysis of relative clauses. This will also give us the opportunity to show how the OT-evaluator differs from the fi lters assumed in the principles-and- parameters approaches. Consider again the relative clauses in (14) and (15), which were accounted for in Filters and Control by an appeal to the Doubly Filled COMP Filter and the recoverability condition on deletion.

(14) (a) the man [[COMP who tha t] I know twho] (b) the man [[COMP who that] I know twho] (c) the man [[COMP who that] I know twho] (d) *the man [[COMP who that] I know twho]

(15) (a) the book [about which that he spoke tabout which] (b) *the book [about which that he spoke tabout which] (c) *the book [about which that he spoke tabout which] (d) *the book [about which that he spoke tabout which]

When we contrast these examples with the French relative clauses in (16) and (17), we see that English and French differ in that the former allows a wider variety of constructions with a bare relative pronoun than the latter. However, when the rela-tive pronoun is embedded in a PP (or an NP), the two languages behave the same.

(16) (a) *l’homme [quii que je connais ti] (b) l’homme [quii que je connais ti] (c) *l’homme [quii que je connais ti] (d) *l’homme [quii que je connais ti]

14 Hans Broekhuis and Ralf Vogel

(17) (a) l’homme [avec quii qu e j’ai dansé ti] (b) *l’homme [avec quii que j’ai dansé ti] (c) *l’homme [avec quii que j’ai dansé ti] (d) *l’homme [avec quii que j’ai dansé ti]

In order to account for the data in (14) to (17), Pesetsky proposed the constraints in (18), which we slightly simplify here for reasons of exposition. Constraint (18a) is simply the recoverability condition on deletion from Chomsky and Lasnik (1977), constraint (18b) is a constraint that expresses that embedded clauses tend to be intro-duced by a complementizer, and (18c) is a constraint that expresses that function words (like complementizers) tend to be left unpronounced.

(18) (a) RECOVERABILITY (REC): a syntac tic unit with semantic content must be pronounced unless it has a suffi ciently local antecedent.

(b) LEFT EDGE (CP): the fi rst leftmost pronounced word in an embedded CP must be the complementizer.

(c) TELEGRAPH (TEL): do not pronounce function words.

The ranking of these constraints will determine the optimal output. In order to see this, it is important to note that LE(CP) in (18b) and TEL in (18c) are in confl ict with each other: the fi rst wants the complementizer to be pronounced, whereas the latter wants it to be deleted. Such confl icts make it possible to account for variation between languages: when we rank these constraints differently, we get languages with different properties. When we assume that LE(CP) outranks TEL, we get a language in which embedded declarative clauses must be introduced by a complementizer. When we assume that TEL outranks LE(CP), we get a language in which embedded declarative clauses are not introduced by a complementizer. When we assume that the two constraints are in a tie (ranked equally high), we get a language in which embedded declarative clauses are optionally introduced by a complementizer. The evaluations can be made visible by means of tableaux. Tableau T1 gives the evalu-ation of embedded declarative clauses with and without a pronounced complemen-tizer in a language with the ranking LE(CP) >> TEL.

T1 No complementizer deletion in embedded declarative clauses

LE(CP) TEL

… [ complementizer …] *

… [ complementizer …] *!

The two asterisks indicate that the constraint in the header of their column is violated. The fi rst candidate, with a pronounced complementizer, violates TEL but this is tolerated because it enables us to satisfy the higher ranked constraint LE(CP). The second candidate, with a deleted complementizer, violates LE(CP), and this is fatal (which is indicated by an exclamation mark) because the fi rst candidate does not violate this constraint. The fi rst candidate is therefore optimal, which is indi-

Introduction 15

cated by means of the pointed fi nger: . The shading of the cells indicates that these cells do not play a role in the evaluation; this convention is mainly for convenience, because it makes it easier to read the tableaux.

Now consider the evaluation of the same candidates in a language with the ranking TEL >> LE(CP), given in T2. Since TEL is now ranked higher than LE(CP), violation of the former is fatal, so that deletion of the complementizer becomes obligatory.

T2 Obligatory complementizer deletion in embedded declarative clauses

TEL LE(CP)

… [ complementizer …] *!

… [ complementizer …] *

Tableau T3 gives the evaluation of a language in which the two constraints are in a tie LE(CP) <> TEL, which is indicated in the tableau by means of a dashed line. Under this ranking, the rankings LE(CP) >> TEL and TEL >> LE(CP) are in a sense simultane-ously active. Therefore we have to read the tie in both directions: when we read the tie from left to right, the violation of LE(CP) is fatal (which is indicated by >), and the fi rst candidate is optimal; when we read the tableau from right to left, the violation of TEL is fatal (which is indicated by <), and the second candidate is optimal. This predicts that deletion of the complementizer is optional in this case.

T3 Optional complementizer deletion in embedded declarative clauses

LE(CP) TEL

… [ complementizer …] <*

… [ complementizer …] *>

Let us now return to the difference between English and French with respect the pronunciation of relative clauses. It is clear that English has the tied ranking LE(CP) <> TEL, given that the complementizer is normally optional in embedded declarative clauses. In French, on the other hand, it is clear that LE(CP) outranks TEL given that the complementizer is obligatory in embedded declarative clauses. Pesetsky (1997) has shown that this also accounts for the differences between the English and French examples in (14) and (16), in which a bare relative pronoun is preposed. Assume that in both languages the constraint RECOVERABILITY outranks the constraints TEL and LE(CP); the ranking of the constraints in (18) are then as given in (19).

(19) (a) French: REC >> LE(CP) >> TE L (b) English: REC >> LE(CP) <> TEL

16 Hans Broekhuis and Ralf Vogel

The evaluation of the French examples in (16) proceeds as in T4. Since the rela-tive pronoun has a local antecedent it is recoverable after deletion, so that all candi-dates satisfy REC. The second candidate is the optimal candidate because it is the only one that does not violate LE(CP); the fact that this candidate violates the lower-ranked constraint TEL is tolerated since this in fact enables the satisfaction of the higher-ranked constraint LE(CP).

T4 Relative clauses with preposed relative pronoun

French REC LE(CP) TEL

l’homme [quii que je connais ti] *!

l’homme [quii que je connais ti] *

l’homme [quii que je connais ti] *!

l’homme [quii que je connais ti] *! *

The evaluation of the English examples is slightly more complex than that of French due to the fact that LE(CP) and TEL are in a tie: we are therefore dealing with two rankings at the same time: REC >> LE(CP) >> TEL and REC >> TEL >> LE(CP). The fi rst ranking is actually the one we also fi nd in French, and we have seen that this results in selection of the second candidate as optimal. Under the second ranking, violation of TEL is fatal, so that the fi rst and third are selected as optimal. As a result, three out of the four candidates are grammatical in English.

T5 Relative clauses with preposed relative pronoun

English REC LE(CP) TEL

the man [whoi that I know ti] *>

the man [whoi that I know ti] <*

the man [whoi that I know ti] *>

the man [whoi that I know ti] *> <*

Next consider the evaluation of the French examples in (17), in which a PP containing a relative pronoun is preposed. Since the preposition is not locally recov-erable, deletion of it leads to a violation of the highest-ranked constraint REC: this excludes the second and the third candidate. Since the two remaining candidates both violate LE(CP), the lowest ranked constraint TEL gets the fi nal say by excluding the fourth candidate. Note that this shows that the ranking LE(CP) >> TEL does not mean that the complementizer is always realized, but that this may depend on other factors; when the complementizer is preceded by some element that must be real-ized, TEL forces the complementizer to delete.

Introduction 17

T6 Relative clauses with preposed PP

French REC LE(CP) TEL

l’homme [avec quii que j’ai dansé ti] *

l’homme [avec quii que j’ai dansé ti] *! *

l’homme [avec quii que j’ai dansé ti] *! *

l’homme [avec quii que j’ai dansé ti] * *!

For the English examples in (15) we get the same result as in French: both the second and the third candidate are excluded by REC, and the fourth candidate is excluded because it is harmonically bound by the fi rst candidate: it has a fatal viola-tion of TEL irrespective of the question whether we read the tie from left to right or from right to left.

T7 Relative clauses with pr eposed PP

English REC LE(CP) TEL

the book [about whichi that he spoke ti] *

the book [about whichi that he spoke ti] *! *

the book [about whichi that he spoke ti] *! *

the book [about whichi that he spoke ti] * *!

2.3 Conclusion

This section has argued that the global architecture of grammar is as given in Figure 1.1, and that the several proposals made within the P&P approach do not differ in this respect from OT. The two frameworks are similar in assuming that we are dealing both with derivations and with evaluations: a generator creates a potentially multi-membered set of expressions S, and an evaluator determines which expres-sions from S are grammatical in a given language L. The OT view on the evaluator seems to be of a more optimistic nature than that of the P&P approaches. The latter consider the evaluator as a more or less random collection of language-specifi c fi lters on the output of narrow syntax. Pesetsky’s work has shown, however, that at least some of the fi lters proposed by Chomsky and Lasnik (1977) can be decomposed into more atomic OT constraints, and Dekkers (1999) has shown this for a number of other constraints/principles like the ECP. Furthermore, since the OT constraints are claimed to be universal, they make precise predictions about the range of language variation that is allowed: Pesetsky, for example, has shown that his proposal is able to account for the differences between English and French relative clause construc-tions, and Broekhuis and Dekkers (2000) and Dekkers (1999) have shown that his proposal can be readily extended to relative constructions in Dutch.

18 Hans Broekhuis and Ralf Vogel

3. Where MP and OT do diff er: d erivat ions a nd e valuations

The previous section has shown that MP and OT assume the same global architec-ture of grammar, but there are, of course, also a number of obvious differences. This subsection will argue, however, that these do not have a principled linguistic moti-vation, but are the result of a more or less accidental difference in focus of attention between the two approaches: MP is mainly concerned with the universal (deriva-tional) aspects of grammar whereas OT-syntax rather focuses on more language-specifi c aspects of grammar. This focus of interest is also refl ected in the research strategies that the two approaches employ: research in MP tends to attribute to the generator CHL as many properties of languages as possible, whereas Research in OT tends to appeal to the evaluator instead. It is therefore not surprising that the empir-ical successes of the two approaches also lie in different areas: MP is especially well equipped to account for the universal properties of languages, but there is no generally accepted view on the way we should account for, or even approach, the many ways in which languages may differ from each other; OT, on the other hand, precisely provides such a general theory of language variation, but since there is no generally accepted theory of the generator, current OT-syntax fails to account for the ‘truly’ universal properties of languages. These differences between MP and OT will be discussed more extensively below.

3.1 Universal properties of language ( the generator)

Both MP and OT-syntax hold the generator responsible for the invariant properties of language: the generator determines what representations are contained in the output, and hence can take part in the evaluation. The two frameworks differ, however, with respect to the extent that the generator is developed, or invoked in the analysis of the linguistic data.

The investigation of the generator (CHL) is considered MP’s core business. It has resulted in a sophisticated, restrictive theory on the nature of the generator. It is assumed that CHL is constituted by a small set of operations that are subject to inviolable conditions that are relatively well understood. Perhaps CHL can be reduced to a single merge operation, which has two incarnations, external and internal merge. As a result of this, the output of CHL is also highly restricted; although it can be a non-singleton set, the differences between the members of this set are very limited in nature, and perhaps only involve the number of movements that occurred; cf. the discussion of Icelandic OS in section 2.1. It seems that analyses that do not invoke fi ltering devices are valued higher in MP than those that do, and, as a result, research tends to focus on those phenomena that can be successfully approached by means of a derivational account, with a concomitant reduction of the empirical scope of the theory, as is clear from the fact already mentioned earlier that Chomsky (1995: §4.7.3) suggests that ‘rearrangement’ phenomena like extraposition, right-node raising, VP-adjunction and scrambling are not part of narrow syntax. It is generally admitted in OT-syntax that the generator is the locus of the ‘truly’ universal properties of language: Grimshaw

Introduction 19

(1997), for example, assumes that the structures formed by the generator conform to some version of X-bar-theory, Pesetsky (1998) and Anderson (2000) adopt some version of generative grammar as the generator, and Bresnan (2000) and Sells (2001) argue in favor of some version of Lexical Functional Grammar. The nature of the generator is, however, not a prominent subject of research; it is rather exceptional for an OT-researcher to account for some phenomenon by taking recourse to the gener-ator given that most research in OT-syntax focuses on the variation that can be found rather than on the universal properties of languages.

Despite the differences in theoretical background (P&P, LFG, etc.), it seems that the view on the generator of many (if not most) OT-syntacticians crucially differs from that of MP-researchers, which becomes especially apparent when we consider the differences in the view on the output of the generator. We have already seen that, although MP allows for non-singleton output sets, MP-researchers generally take it for granted that this set is very small and that differences between the members of this set are limited in type, perhaps confi ned to differences in movement. In OT, on the other hand, it is generally maintained that the output of the generator can in prin-ciple be infi nitely large, and that the members of the set may differ in a wide variety of ways. This suggests that the generator is generally taken to contain a larger set of operations in OT than is assumed in MP, and that these operations are probably confi ned in a less strict manner than the operations assumed in MP.

As a result of this different view on the generator, MP and OT tend to provide entirely different explanations for similar phenomena, the former taking recourse mainly to properties of the generator and the latter to those of the evaluator. This state of affairs seems to strengthen the widely accepted view that we are dealing with two competing and essentially incompatible frameworks. However, it can also be assessed differently, and more positively. Since it is not a priori given whether a certain phenomenon should be accounted for by appealing to the generator or the evaluator, it is important to develop alternative analyses that can subsequently be compared and evaluated; the fact that in some domains competing MP- and OT-analyses are available therefore does not mean in itself that we are dealing with competing or confl icting theories.

3.2 Variation (the evaluator)

One of the main concerns of both MP and OT is cross-linguistic variation. However, the way they approach this problem is entirely different – at least, at fi rst sight. Let us start with discussing the way MP approaches the issue. Language variation is assumed to arise as a result of additional constraints on the application of the otherwise universal generator (CHL). The generator can basically perform two oper-ations: external and internal merge. Let us provisionally adopt the standard assump-tion in MP that external merge is indispensable given that it is needed in order to assemble lexical items into semantically interpretable structures, for example, by the saturating the thematic roles of a given lexical head. Despite the fact that internal merge (movement) may have certain semantic implications, it is not essential in the

20 Hans Broekhuis and Ralf Vogel

creation of semantically interpretable structures, so that we expect to fi nd language variation in this domain. Note that since MP is mainly concerned with narrow syntax it mainly studies differences between languages that are somehow related to move-ment: variation in other domains is attributed to other modules (like PF), and is generally not discussed any further.

In early MP, the locus of variation between languages is solely attributed to the lexicon. Differences in the displacement property of languages are due to differ-ences in the ‘strength’ property of the morpho-syntactic features that trigger move-ment: strong features trigger overt movement, whereas the weak features allow covert movement (which is favored by Procrastinate). In the more recent Agree-based theories, which reject the idea of covert movement, the core idea is preserved by assuming that movement only takes place if a functional head F contains an EPP-feature, which requires that the specifi er of F be present. Under this view, the task of the language learner is to determine whether the functional head F has a weak or strong feature, or, alternatively, whether it has an EPP-feature, and to store this infor-mation in the lexicon.

The scope of OT goes much beyond the displacement property of languages: in principle, all (phonological, syntactic, semantic, pragmatic, etc.) properties can be fruitfully investigated, as long as one can plausibly postulate constraints bearing on the phenomenon in question. As we have already seen variation between languages is attributed to the evaluator, more specifi cally to the differences in ranking of the otherwise universal constraints. Under this view, the task of the language learner is to determine the constraint ranking (as well as the lexicon) of the language.

The discussion above seems to reveal another important difference between MP and OT: in the former, cross-linguistic variation is solely due to differences in lexical specifi cations, whereas in the latter it is rather due to the ranking of the universal constraints. This is indeed the case when we compare early MP with OT-syntax, but it no longer holds when we compare the most recent Minimalist Inquiry framework and OT-syntax. The early MP thesis that the sole locus of cross-linguistic variation is the lexicon runs into severe problems when we consider variation within a single language, because it predicts that languages cannot have ‘optional’ movement, that is, movements that occur only under well-defi ned semantic or phonological condi-tions. One example of this type of movement is Icelandic OS (already discussed in section 2.1), which can only apply when the object is part of the presupposition of the clause (cf. (6)), and when it does not cross the verb (cf. (7)) or other v*P-internal material. This kind of optionality cannot arise under the early MP thesis because the postulation of feature strength or an EPP-feature gives rise of to a very rigid system: when a feature is strong/an EPP-feature is present, movement must apply; when a feature is weak/an EPP-feature is not present, movement is blocked by Procrastinate.

We have already seen that this problem has led Chomky (2001) to assume that the EPP-feature, which forces movement, is optionally present. In order to avoid circu-larity, the choice must be made sensitive to external factors like the semantic and phonological conditions imposed on the pertinent movement, and this is precisely what Chomsky did in his account of OS in Icelandic in (10), repeated below as (20): as we have seen, the language-specifi c statement in (20c), in tandem with the

Introduction 21

universal principles in (20a,b), precisely derives the circumstances under which Icelandic OS applies.

(20) (a) v* is assigned an EPP-fea ture only if that has an effect on outcome. (b) The EPP position is assigned INT. (c) At the phonological border of v*P, XP is assigned INT.

Chomsky (2001: 36) presents clause (20c) as a parameter that distinguishes OS from non-OS languages. French, for example, has verb movement to I, but never-theless OS does not apply. This can be accounted for by assuming that (20c) does not hold for French. As a result, the interpretation INT can be assigned to the object when it is at the phonological border of v*P; as a result, movement of the object to the EPP-position is not needed and assignment of an EPP-feature to v* is consequently blocked by (20a).

It seems, however, that (20c) is unlike the parameters of the earlier P&P frame-work in that it is not binary, because it is not the case that languages can be straight-forwardly divided between OS and non-OS languages. This will become clear when we consider the Danish examples in (21) and (22), taken from Vikner (1994: 502); The examples in (21) show that Danish differs from Icelandic in that it does not have OS of non-pronominal DPs, whereas the examples in (22) show that it does have OS of weak pronouns.

(21) (a) Hvorfor læste studenterne ikke artiklen? why read the students not the article (b) *Hvorfor læste studentene artikleni ikke ti?

(22) (a) Hvorfor læste studenterne d eni ikke ti? why read the students it not (b) *Hvorfor læste studenterne ikke den?

This can be accounted for by assuming that clause (20c) must be further refi ned as in (20c). This clause correctly expresses: (a) that non-pronominal DPs that are part of the presupposition of the clause (= INT) must undergo OS in Icelandic, but not in Danish or the Romance languages; and (b) that defi nite pronouns (which are assigned INT by defi nition) must undergo OS in Icelandic and Danish but not in the Romance languages.

(20) (c) At the phonological border of v*P, XP is assigned INT (i) XP = DP (Icelandic) (ii) XP = defi nite pronoun (Danish) (iii) XP = (Romance)

What we want to stress here is that the adoption of language specifi c statements like (20c) or (20c) is a radical breaks with the early MP thesis that the sole locus of cross-linguistic variation is the lexicon. Since these statements essentially func-tion as language-specifi c fi lters on the output of CHL, linguistic variation should

22 Hans Broekhuis and Ralf Vogel

be attributed to the evaluator in the model in Figure 1.2, and not to the lexicon. In fact, it seems that Chomsky’s proposal makes it possible to eliminate the EPP-features entirely: when we assume that movement is subject to Last Resort but applies optionally, we could simply replace clause (20a) by the claim that movement is possible only if it has an effect on the outcome. This would make it possible to attribute cross-linguistic language variation entirely to the evaluator, just like in OT. In (23) we attempt to rephrase Chomsky’s proposal such that reference to the notion of EPP-feature becomes superfl uous.

(23) (a) Movement is possible only if it h as an effect on outcome. (b) The derived object position is assigned INT. (c) At the phonological border of v*P, XP is assigned INT. (i) XP = DP (Icelandic) (ii) XP = defi nite pronoun (Danish) (iii) XP = (Romance)

3.3 Conclusion

Since we have seen that MP and OT assume more or less the same global organi-zation of grammar, we may conclude that the differences in the research strategies of MP and OT are somewhat accidental: as far as we can see, there are no theory-internal reasons for these frameworks to limit their investigation to respectively the generator or the evaluator. The fact that MP and OT occasionally provide alternative analyses for similar data as a result of these differences in research strategy does not follow from insurmountable theoretical differences between the two frameworks either, but simply refl ects the fact that it is not a priori given whether a certain phenomenon belongs to the computational system or to the fi lter component of core grammar. Early MP and OT-syntax do seem to adopt confl icting views on the nature of variation between languages: the former adopts the thesis that language variation can be reduced to differences in the feature specifi cations of the lexical elements (feature strength/EPP-features), whereas the latter assumes that language variation is due to the evaluator, that is, to differences in constraint rankings. In Chomsky’s current Minimalist Inquiry framework, however, the early MP thesis has been dropped: language variation is (also) attributed to parameters like (23c), which essentially function as language-specifi c fi lters on the output of CHL. Current MP and OT therefore both attribute language variation to the evaluator, and the main differ-ence between MP and OT boils down to the question whether the evaluator appeals to output fi lters or to ranked constraints.

4. Organization of the book

The discussion above has shown that MP and OT-syntax are actually much more alike than is generally assumed or one would think at fi rst sight. The least one can

Introduction 23

say is that they assume a similar global architecture of the grammar, and thus face a number of similar questions, like:

(24) (a) What are the defi ning propert ies of the generator? (b) What are the defi ning properties of the evaluator? (c) What is the division of labor between the generator and the evaluator? (d) What is the role of the articulatory-perceptual and the conceptual-intentional

system?

We therefore have to ask ourselves whether it is still justifi ed to consider MP and OT different, divergent programs, or whether it is possible to combine the best results of these programs into a single theory of grammar. The studies that will follow all discuss these questions from different perspectives and thus provide a collection of possible answers to these questions. Some studies in fact go beyond this and show alternative ways in which the output of the derivational system can be fi ltered. This introduction will not summarize the individual studies or address their contents in detail; for this we refer to the abstracts and the studies themselves. However, we will conclude this introduction by giving a brief discussion of the organization of the book as a whole and the grouping of the individual studies.

4.1 Part I: combining MP with an OT-evaluation

The fi rst group of studies presents work within the so-called derivation-and- evaluation framework, which explicitly combines MP and OT into a hybrid system: more specifi cally, it is argued that some version of the computational system of human language CHL from MP functions as a generator that creates a restricted set of potentially well-formed expressions, which are subsequently evaluated in an optimally theoretic fashion by means of limited number of violable constraints. The chapter by Hans Broekhuis introduces this framework and illustrates it by means of a topic that was also extensively discussed in this introduction: Object Shift. He shows that Chomsky’s proposal can be readily rephrased in optimality-theoretic terms and that this has a wide range of empirical consequences. The chapter by Gema Chocano and Mike Putnam discusses a number of restrictions on the licensing of parasitic gaps that are problematic in a purely derivational framework like MP but which fall out quite naturally under the hybrid approach by adding a single constraint to the inventory proposed by Broekhuis. Martin Salzmann discusses the dialectal and intra-speaker variation that can be found with dative resumption in relative clauses, that is, the use of a clause internal dative pronoun instead of a trace when the antecedent of the relative clause corresponds to the dative argument of the verb. Salzmann argues that the range of variation shows that locality is an inviolable condition on movement, but that the standard version of MP is neverthe-less ill-equipped to handle the attested variation; he therefore concludes that hybrid systems like the derivation-and-evaluation framework are optimal for expressing the correct generalizations.

24 Hans Broekhuis and Ralf Vogel

4.2 Part II: local and global optimization

As we have seen above one of the conspicuous differences between MP and tradi-tional OT is that the former focuses on the derivation (the generator) whereas the latter focuses on the output representations of the generator (the evaluator). The fi rst two studies collected in Part II show that this is by no means a necessary differ-ence between MP and OT and that it is readily possible to formulate derivational versions of OT.

The fi rst example of such a derivational version of OT is Harmonic Serialism (HS), which is illustrated by John McCarthy and Kathryn Pruitt on the basis of stress assignment. According to HS, the output of the generator is evaluated in a step-wise fashion by means of a Gen-Eval loop: at each point in the derivation the output of the generator is evaluated by the evaluator, after which the optimal output is sent back to the generator. This results in a process of local optimization that continues until the input and the optimal output representations are identical; this is the point of conver-gence. McCarthy and Pruitt explicitly compare the derivation of metrical structure in HS to the structure building operations found in MP, and claim that ‘both theories seek to explain the derivation of complex structures by deriving them via repeated application of simple operations under the control of an optimizing grammar.’

The chapter by Fabian Heck and Gereon Müller seems to fi t seamlessly in this view given that they argue that the application of the structure building operations of MP are constrained by a local optimization in a way that comes very close to HS. They distinguish structure-building features, which trigger Merge, and probe features, which trigger Agree, and assume that the order of saturation of these features is determined in a local, stepwise fashion. Empirical evidence in favor of local opti-mization is provided by cross-linguistic differences in case assignment to internal and external arguments (that is, the difference between nominative-accusative and ergative-absolutive languages), agreement patterns in German DPs with prenominal dative possessors, the use of the German expletive es, and VP-topicalization. Heck and Müller suggest that many analyses that involve larger domains can be rephrased in terms of local optimization, but admit that there may be certain analyses that may require larger optimization domains.

Although the contributions by McCarthy and Pruitt and Heck and Müller show that local optimization has clear advantages when it comes to the application of the operations of the generator, there are cases where global optimization seems more suitable. Heck and Müller, for example, predict a dichotomy between nominative-accusative and ergative-absolutive languages, but Ellen Woolford shows that the distinction is not always clear-cut given that in some languages the expression of ergative case depends on certain contextual properties of the construction as a whole. For example, in Hindi and Nepali ergative case is only used in, respectively, perfec-tive and individual-level contexts; since this information is probably not available or accessible during the derivation it seems less likely that this can be accounted for by means of local optimization.

Another case that requires global optimization involves the order restrictions on Scandinavian object shift constrictions (see Section 3 above). This is again

Introduction 25

underlined by the discussion of object shift in Scandinavian remnant topicaliza-tion constructions by Eva Engels and Sten Vikner, who show that the optimization domain must be at least as large as CP. Their contribution also contains a discussion of the cyclic linearization approach by Fox and Pesetsky (2005), and they show that this approach is less well equipped to account for the set of data they discuss than their OT-approach.

The studies in this section suggest that we may need to postulate both local and global optimization. This need not be frowned upon with suspicion as this actu-ally refl ects the current distinction found within MP between local condition on the operations of the computational system CHL and the set of conditions imposed by the articulatory-perceptual and the conceptual-intentional systems on output repre-sentations. It simply shows that OT-evaluations may be pervasive in the grammar in the sense that both types of restrictions may be rephrasable in terms of violable OT-constraints.

4.3 Part III: optimal design, economy and last resort in OT

Chomsky has stressed at a number of occasions that the name minimalist program is less felicitous given that minimalist considerations are a defi ning part of any scien-tifi c enterprise; minimalist concerns are therefore expected to play an important role in the development of the more traditional OT-approaches as well. The studies collected in this part of the book are good examples of this.

Vieri Samek-Lodovici argues that when it comes to cross-linguistic variation OT meets the requirement of optimal design better than the traditional versions of MP developed in the 1990s. He shows that under the assumption that no ranking of the universal, potentially confl icting (hence violable) constraints is inherently superior to any other, language variation is a predicted outcome of OT, whereas under the postulation of universal inviolable (hence non-confl icting) conditions language vari-ation requires language-specifi c stipulations like Chomsky’s (1993) postulation of weak/strong features on certain lexical items. Samek-Lodovici further points out that the existence of confl icting constraints is implied by the postulation of bare output conditions: since the sensory-motor and the conceptual-intentional system serve largely independent goals, there is no reason to exclude the possibility of confl icting interface constraints.

One way in which OT and minimalism can coexist and complement each other lies in OT’s potential to model the interfaces between syntax, semantics and phonology/phonetics. A leading idea of minimalism is to reduce syntax to what is ultimately necessary to fulfi ll the needs of these interfaces. Therefore, the more elaborate the interfaces are constructed, the simpler the syntactic generator might be construable. Ralf Vogel concludes in his chapter that only very few of the specifi cally minimalist properties of the syntactic generator are necessary for OT’s syntax generator, when one exploits OT as interface theory as much as possible. The conception of an OT grammar that he argues for organizes the mapping between semantic, syntactic and phonolog-ical/phonetic representations using violable and confl icting mapping constraints. The

26 Hans Broekhuis and Ralf Vogel

core concept of OT that is relevant here is faithfulness, formulated in a corresponding theoretic way. Vogel shows that from this perspective the optimal syntactic represen-tations are not necessarily the most economical ones, but rather those that correspond best to semantic and phonological/phonetic representations. He further argues that OT’s notion of markedness is more adequate than MP’s notion of economy. Unmarked syntactic structures can be seen as part of such maximally isomorphic mappings.

The notion of economy in MP is often linked to the notion of last resort: a certain operation can only be used when it is needed to arrive at a converging derivation. In early MP, for example, movement was claimed to be possible only when it serves to check and eliminate an uninterpretable feature of a certain sort. Jane Grimshaw criti-cizes this use of the notion given that we could simply remove the ‘last resort’ concept and state that movement is only possible when a certain feature is present (as is indeed assumed in the later versions of MP that assume optional EPP-features; see our earlier discussion of Chomsky’s account of object shift). Grimshaw further claims that the notion of last resort can only receive a coherent interpretation in theories of optimiza-tion with constraint interaction. In fact, it is claimed that the notion is in fact entailed by such theories and that the use of any grammatical device is the result of last resort: it is the best that can be done in a particular confi guration given a certain constraint ranking. This is illustrated by means of the choice between V-to-T/C, do-support and free tense (tense not supported by a verb) in a variety of constructions and languages. The discussion shows that, contrary to popular belief, do-support is not language-specifi c but arises in different circumstances in different languages.

4.4 Part IV: the role of the interpretative components

Standard MP assumes that the interpretative components (PF and LF) impose certain conditions on the output of the generator, and we have further seen that such inter-face conditions can be readily expressed in an OT-fashion by means of ranked, violable constraints. The two studies in this part of the book propose alternative ways in which the interpretative component may affect the output of the generator.

Like Samek-Lodovici, Hedde Zeijlstra argues that the conditions imposed by the sensory-motor and the conceptual-intentional system on the output representations of the generator are necessarily in confl ict and therefore (at least partly) violable. This gives rise to a tension that is solved by different languages in different ways, with language variation as a result. Zeijlstra postulates an inviolable principle of Full Legibility, which requires that all elements in the output be legible at the level of LF and PF, but which differs from Chomsky’s (1995) Full Interpretation in that it allows legible but uninterpretable elements to be present; as a result, Full Legibility can be satisfi ed in more than one way. However, given that uninterpretable elements do not facilitate legibility, their number should be reduced as much as possible, and Full Legibility thus invokes simplicity measures that disfavor the presence of such elements. Since the set of legible elements differs for the level of represen-tation (PF or LF) we are dealing with, the simplicity measures imposed on the output of the computational system may be in confl ict: reduction of uninterpretable

Introduction 27

elements at LF may result in an increase of uninterpretable elements at PF, and vice versa. Zeijlstra claims that this may result in more than one ‘optimal solution’ and that languages may select different solutions as the grammatical option. He claims that the actual choice is determined by the ancestry/acquisition of the language in question: the simplicity measures select the simplest grammar compatible with the target language (which may also account for the fact that some languages seem to select a suboptimal solution). Zeijlstra thus agrees with traditional OT that the inter-pretative components impose confl icting violable constraints on the output repre-sentations of the generator without, however, using the OT-formalism of constraint ranking in his account of the selection of the grammatical candidates for a given language L.

Kleanthes Grohmann argues that the PF-component determines how the copies of movement are spelled out. The basic hypothesis is, however, that this is done by means of inviolable, universal conditions on the output representations: he divides the clause in three mutually exclusive domains (in which respectively the thematic, agreement and discourse information is expressed). He further shows that when the moved element and its copy are within the same domain, the latter must be phoneti-cally expressed; in other confi gurations an ‘elsewhere’ condition requires deletion of the copy. Earlier work has shown that this proposal may account for a wide range of phenomena, but Grohmann also shows that there is a small range of facts in which the copy is unexpectedly spelled out due to the intervention of other ‘independent constraints of the grammar’; cf. his (41b). It seems that such cases may be a good testing ground for some of the proposals discussed in this book.

Acknowledgments

The majority of studies collected in this book were papers presented at the workshops on Descriptive and Empirical Adequacy held in Berlin (December 2005) and Leiden (February 2008). We like to thank everyone who made these events possible as well as the contributors to the present book. We especially thank our assistants Ann-Christin Broschinski, Anna Kutscher, and George Moore, who did an amazing job in the fi nal preparations of the manuscripts. Ben Hermans helped us with editing the manuscript by John McCarthy and Kathryn Pruitt.

We hope that this book will encourage people who are currently exclusively working in MP or OT to broaden their vision by also considering the possibilities that the alternative framework offers, and that this will stimulate further fruitful debate among proponents of the two frameworks. The research of Hans Broekhuis for this book was partly funded by a so-called VIDI-grant (2003-8) from the Netherlands Organization of Scientifi c Research (NWO).

References

Anderson, S. R. (2000) Towards an optimal account of second-position phenomena. In J. Dekkers, F. Van der Leeuw and J. Van de Weijer (eds) Optimality Theory: Phonology, Syntax and Acquisition, 302–33. Oxford: Oxford University Press.

Bobaljik, J. D. (2002) A-chains at the PF-interface: Copies and ‘covert’ movement. Natural Language & Linguistic Theory 20: 197–267.

Bresnan, J. (2000) Optimal syntax. In J. Dekkers, F. Van der Leeuw and J. Van de Weijer (eds) Optimality Theory: Phonology, Syntax and Acquisition, 334–85. Oxford: Oxford University Press.

28 Hans Broekhuis and Ralf Vogel

Broekhuis, H. (2008) Derivations and Evaluations: Object Shift in the Germanic Languages. Berlin/New York: Mouton de Gruyter.

Broekhuis, H. and Dekkers, D. (2000) The minimalist program and optimality theory: Derivations and evaluations. In J. Dekkers, F. Van der Leeuw and J. Van de Weijer (eds) Optimality Theory: Phonology, Syntax and Acquisition, 386–422. Oxford/New York: Oxford University Press.

Chomsky, N. (1981) Lectures on Government and Binding. Dordrecht: Foris.Chomsky, N. (1986) Barriers. Cambridge, MA: MIT Press.Chomsky, N. (1991) Some notes on economy of derivation and representation. In R. Freidin (ed.)

Principles and Parameters in Comparative Syntax, 417–54. Cambridge, MA: MIT Press.Chomsky, N. (1995) The Minimalist Program. Cambridge, MA: MIT Press.Chomsky, N. (2000) Minimalist inquiries: The framework. In R. Martin, D. Michaels and J. Uriagereka

(eds) Step by Step. Essays on Minimalist Syntax in Honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press.

Chomsky, N. (2001) Derivation by phase. In M. Kenstowicz (ed.) Ken Hale. A Life in Language, 1–52. Cambridge, MA: MIT Press.

Chomsky, N. and Lasnik, H. (1977) Filters and control. Linguistic Inquiry 8: 425–504.Dekkers, J. (1999) Derivations & Evaluations. On the Syntax of Subjects and Complementizers, Doctoral

dissertation. University of Amsterdam/HIL.Fox, D. and Pesetsky, D. (2005) Cyclic linearization of syntactic structure. Theoretical Linguistics 31:

1–45.Frampton, J. and Gutmann, S. (2002) Crash-proof syntax. In S. D. Epstein and T. D.Seely (eds)

Derivations and Explanations in the Minimalist Program, 90–105. Malden, MA and Oxford: Blackwell Publishing.

Grimshaw, J. (1997) Projection, heads and optimality. Linguistic Inquiry 28: 373–422.Groat, E. and O’Neil, J. (1996) Spell-Out at the LF interface. In W. Abraham, S. D. Epstein and H.

Thráinsson (eds) Minimal Ideas. Syntactic Studies in the Minimalist Framework, 113–39. Amsterdam and Philadelphia, PA: John Benjamins.

Holmberg, A. (1999) Remarks on Holmberg’s generalization. Studia Linguistica 53: 1–39.Kayne, R. (1994) The Antisymmetry of Syntax. Cambridge, MA: MIT Press.Müller, G. (2000) Shape conservation and remnant movement. In A. Hirotani, N. Hall Coetzee and J.-Y.

Kim (eds) Proceedings of NELS 30, 525–39. Amherst, MA: GLSA.Müller, G. (2001) Order preservation, parallel movement, and the emergence of the unmarked. In G.

Legendre, J. Grimshaw and S. Vikner (eds) Optimality-theoretic Syntax, 113–42. Cambridge, MA and London: MIT Press/MITWPL.

Pesetsky, D. (1997) Optimality theory and syntax: Movement and pronunciation. In D. Archangeli and T. Langendoen (eds) Optimality Theory, 134–70. Malden, MA and Oxford: Blackwell.

Pesetsky, D. (1998) Some optimality principles of sentence pronunciation. In P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis and D. Pesetsky (eds) Is the Best Good Enough?, 337–83. Cambridge, MA and London: MIT Press/MITWPL.

Putnam, M. (2010) Exploring Crash-proof Grammars. Amsterdam: John Benjamins.Rizzi, L. (1996) Residual Verb Second and the Wh criterion. In A. Belletti and L. Rizzi (eds) Parameters

and Functional Heads. Essays in Comparative Syntax, 63–90. Oxford and New York: Oxford University Press.

Sells, P. (2001) Structure Alignment and Optimality in Swedish. Stanford, CA: CSLI Publications.Vikner, S. (1994) Scandinavian object shift and West Germanic scrambling. In N. Corver and H. van

Riemsdijk (eds) Studies on Scrambling. Movement and Non-movement Approaches to Free Word-Order Phenomena, 487–517. Berlin and New York: Mouton de Gruyter.

Vogel, R. (2004) Correspondence in OT syntax and minimal link effects. In G. Fanselow, A. Stepanov and R. Vogel (eds) Minimality Effects in Syntax, 401–42. Berlin: Mouton de Gruyter.

Vogel, R. (2006) The simple generator. In H. Broekhuis and R. Vogel (eds) Optimality Theory and Minimalism: A Possible Convergence? Linguistics in Potsdam 25, 99–136. Potsdam: University of Potsdam: http://www.ling.uni-potsdam.de/lip.

Woolford, E. (2007) Case locality: Pure domains and object shift. Lingua 117: 1591–616.